RCA Benchmark Packs
Smaller paid packs for QA, vendor evaluation, and pilots
Pre-built subsets of the RCA Insurance and Medical libraries, scoped for fast review or procurement evaluation.
The current pack menu
Each pack is a small, paid, curated subset designed for a specific evaluation use case. Smaller than a full library, larger than a preview. New pack types are added as customer use cases emerge.
Insurance QA Sprint Pack
- 10 complete commercial P&C submission packs
- Ground truth, bboxes, scanned variants, and engineered red flag summary
- 30-minute handover call to walk through the schema and red flag inventory
- AUD $2,500 fixed price
- 48 to 72 hour delivery from order confirmation
- Designed for QA-ing an existing extraction pipeline against a controlled, varied input set
Insurance Procurement Pack
- 5 to 10 packs scoped for vendor bake-off
- All vendors get the same PDFs and the same ground truth
- You score them against the same baseline
- Pricing on request, depending on scope and number of vendors
Medical Review Pack
- 25 to 35 representative medical documents across the major document types
- Designed for a five-minute review path documented in the pack README
- Free for qualified prospects
- Paid extension available if you want a larger review
Medical Pilot Pack
- 100 to 200 documents curated against your specific use case (discharge summary heavy, or pathology heavy, or a specific clinical specialty)
- Priced per scope
- Designed to feed an internal pilot before committing to a full QA library or a vendor
What every Benchmark Pack ships with
- README_START_HERE.md with the recommended review path
- license_summary.md stating what you can and cannot do with the files
- validation_summary.md confirming PDF / ground truth / bbox / scan integrity
- file_list.csv mapping each PDF to its document_id, document_type, template_family, style_profile
- ground_truth_sample.csv and ground_truth_sample.jsonl
- bboxes_sample.jsonl with labelled bboxes (per-claim row and per-location row for insurance)
- manifest_sample.json summarising pack composition
- Insurance packs additionally ship red_flags_summary.csv with where_to_review pointers
Why packs and not just libraries
A library is for ML training and bulk QA. A Benchmark Pack is for a specific human-led evaluation: a review, a vendor bake-off, a pre-rollout QA, a stakeholder demo. Smaller, more curated, faster to consume, and shipped with a recommended review path.
If your use case is "we want to evaluate three vendors against the same baseline", a pack is the right unit. If your use case is "we need 500 labelled discharge summaries to train a model", a library is the right unit.
Pricing summary
| Pack | Price | Delivery |
|---|---|---|
| Insurance 2-pack preview | Free | Same day on request |
| Insurance QA Sprint Pack | AUD $2,500 | 48 to 72 hours |
| Insurance Procurement Pack | On request | Scoped per order |
| Medical 25-doc review pack | Free for qualified prospects | Same day on request |
| Medical Pilot Pack | On request | Scoped per order |
Order the Insurance QA Sprint Pack
10 complete submission packs. Engineered red flags. AUD $2,500. 48 to 72 hour delivery. 30-minute handover call.